SI-7984 Issue unchecked warning for type aliases

[retronym]

The unchecked nature of the type test was already noticed,
but because the type alias tripped up in rendering a sensible
message CheckabilityChecker chose to remain silent, other
than under -Xdev.

This commit:

• Dealiases pattern types before launching the CheckabilityChecker
• Sharpens the error messages to explain that the dealiased type
  is the expansion of the alias.

[jozic]

that was fast ) thanks Jason!

[retronym]

I'll fix the test failures tomorrow morning.

[retronym]

Review by @adriaanm.

I'm tossing up whether or not to suppress the longer error messages (X[A] (underlying of Y[B])) for "direct" aliases like scala.Set. Probably should. I can't find existing code to check for such an alias, but I suppose it won't be too tricky to write.

[adriaanm]

LGTM -- I don't know of any utility code lying around for detecting straight shot aliases